BERTopic It! The Innovative Approach of Applying BERTopic for Text Analysis
(A2026-133506)

AUTHORS

Yicheng Dai, University of Leeds/PGR student; Boshuo Guo, the University of Leeds

ABSTRACT

Topic model serves as an important analytical tool with extensive applications in marketing research. A representative is BERTopic, a new topic model based on advanced technologies. However, the traditional approach of applying this topic model for text analysis has several limitations. Therefore, this study aims to develop an innovative approach to address these issues. To demonstrate the practicality of this approach, we applied it to perform text analysis on an interdisciplinary marketing field. To present the findings of text analysis, we considered a cross-visualization analyzing approach. The results of this study indicate that our applying approach reduces time-consuming in parameter tuning by replacing manual adjustment with machine optimization, and minimizes subjective bias in topic model result evaluation by introducing mathematical coefficients as criteria. Our analyzing approach effectively uncovers insights by employing a holistic analytical perspective instead of a single-dimensional analytical perspective. The managerial implications of this study include three aspects. It is applicable to the text analysis of concise and logically structured content, it can perform independently of hardware constraints, and it provides a comprehensive analysis of diverse types of data, supporting strategic decision-making.